AI & Deep Learning

The use of Artificial Intelligence and Deep Learning is a fundamentally different way of solving problems than traditional applications. In AI we let powerful computers analyse large amount of data to find patterns and connections in the dataset. This process is called training, and it is performed many times over the same dataset, with new pieces of information being learned with every pass. The information is stored in a database, which new data samples can be compared to. This process is called inferencing.

Training of Deep Learning AI models are extremely compute intensive, and require large amount of calculations to be done in parallel. This is a task perfectly suited for GPUs and AI training therefore benefit greatly from GPU acceleration. Inferencing is much less compute demanding than training but can be very time critical. For instance, collision avoidance systems in cars have only a fraction of a second to decide whether to apply the brakes when an object appears in traffic. Many inference tasks therefore also benefit significantly from GPU acceleration.

Some examples of everyday task that use Deep Learning and AI:

  • Picture recognition and image classification, for instance when you search for pictures on Google
  • Analysing speech for voice command systems
  • Driving assistance technologies
  • Fraud detection in financial transactions

Key features you need to get started with AI:

  • Relevant data to analyse – the more the better!
  • Data science personnel – either in house or as hired consultants.
  • Deep learning software – both open source and commercial packages are available. Dedicated AI systems from Nextron come preinstalled with AI software.
  • Lots of compute power. Nextron can deliver a wide selection of AI optimized systems ranging from personal workstations to multi-rack supercomputer clusters!

Further information:

NVIDIA’s introduction to AI and Deep Learning: https://developer.nvidia.com/deep-learning

NVIDIA DGX product family – dedicated AI systems with full software stack and professional support: NVIDIA DGX

AI & Deep Learning